US5093581A

US5093581A - Circuitry for generating pulses of variable widths from binary input data

Info

Publication number: US5093581A
Application number: US07/620,684
Authority: US
Inventors: George R. Briggs; Roger G. Stewart
Original assignee: Thomson SA
Current assignee: Technicolor SA
Priority date: 1990-12-03
Filing date: 1990-12-03
Publication date: 1992-03-03
Anticipated expiration: 2010-12-03
Also published as: DE69122083D1; JPH05504213A; WO1992009187A1; JP3215108B2; EP0513330B1; DE69122083T2; EP0513330A1

Abstract

A variable width pulse generator includes a plurality of logic stages which are coupled in cascade. Each stage is arranged to select one of a plurality of clock signals of differing phase applied thereto responsive to particular bits of a data word defining the pulse width. All of the stages are initially disabled by a precharge pulse occurring at the beginning of each variable pulse interval. The successive stages are enabled by the occurrence of a clock pulse of a clock signal selected by the previous stage. The lastmost stage provides an output corresponding to the variable width pulse.

Description

This invention relates to circuitry for generating control pulses having pulse widths proportional to for example binary numbers applied to such circuitry.

BACKGROUND OF THE INVENTION

U.S. Pat. Nos. 4,742,346 and 4,766,430 issued to Gillette (incorporated herein by reference) illustrate a liquid crystal display (LCD) device incorporating signal drive circuitry integrated together with the LCD elements on a common substrate. Included in the drive circuitry are a plurality of programmable counter circuits. Binary values representative of image brightness are applied to these counters which generate pulses having durations proportional to the binary values. The pulse durations are subsequently converted to potential amplitudes for application to individual display elements.

Consider that the pulses are to represent eight bit binary values, and the longest pulse is approximately equal to the active portion of one horizontal line of video signal, that is approximately 50 μsec. To satisfy these constraints the counter must count at a rate of approximately 5 MHz, e.g., the reciprocal of 50/256 μsec. Now if the counter circuitry is to be realized using amorphous silicon (aSi) transistors as is desirable for economic reasons, this counting rate tends to be too high to be supported by such circuitry. Secondly, programmable counters tend to be relatively complicated, requiring significant numbers of active devices.

SUMMARY OF THE INVENTION

The present invention includes circuitry for generating variable width pulses responsive to binary input values, which circuitry tends to be less complex than programmable counter circuitry and which may operate at lower clock frequencies than programmable counters.

The variable pulse width system includes a plurality of cascade connected stages, each stage for processing one bit of the binary value representing the output pulse width. Each stage includes an output circuit which is precharged to a predetermined state at the beginning of each pulse period. A pair of clock signals φAn, φBn having different phases are applied to gating circuitry which passes one (φAn) or the other (φBn) of the clock signals depending upon whether the applied data bit is a logic zero or a logic one state. The gating circuitry is coupled to the output circuitry of the stage to reset the output potential to a state opposite said predetermined state in response to the clock signal that is passed by the gating circuitry.

The stages are cascaded in order to bit significance such that the most significant bit (MSB) of the binary value is processed first and the least significant bit is processed last. Each successive stage is enabled by the output signal of the previous stage changing state. Output signal from the least significant bit stage represents the variable width pulse.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a block diagram of the variable pulse width generator embodying the present invention.

FIG. 2 is a logic diagram of exemplary circuitry which may be implemented for the respective single bit variable pulse width generating stages of FIG. 1.

FIG. 3 shows potential waveforms useful for describing the FIG. 2 circuitry.

FIG. 4 is a logic schematic diagram showing the interconnection of two of the single bit stages illustrated in FIG. 2.

FIG. 5 is a waveform diagram illustrating the functionality of the FIG. 4 circuitry.

FIG. 6 is a waveform diagram illustrating the requisite clock waveforms for a four bit system.

FIG. 7 is a schematic diagram of an alternative single bit variable pulse generating stage embodying the present invention.

FIGS. 8, 9 and 10 illustrate waveforms descriptive of the operation of the FIG. 7 circuitry.

FIGS. 11 and 12 illustrate clock waveforms for a three bit system utilizing single bit stages of the type illustrated in FIG. 7.

DETAILED DESCRIPTION

FIG. 1 shows in general block form the variable pulse width generator which includes a plurality of single bit stages 90. The single bit stages are connected in cascade for respectively processing digital data bits D1 through Dn, where n is any integer. The single bit stages 90 are each clocked by separate pairs of clock signals φAn, φBn, provided by a clock pulse generator 91.

Each of the cascade connected processing stages has a data bit input terminal for applying one bit of a data word representing the pulse duration; an output terminal and a start pulse input terminal. The start pulse input terminal of each successive processing stage in the cascade connection is coupled to the output terminal of the preceding stage. An externally generated start pulse is applied to the start pulse input terminal of the stage to which the MSB is applied. The data bits are decoded in the order of their bit significance, with each stage being successively enabled by an output transition provided by the next more significant bit stage, which transition corresponds to a predetermined transition of one of the clock phases φAn, φBn applied to that stage. The stage that processes the least significant bit provides an output signal, V_D which is a pulse of duration that represents the magnitude of the data word applied to all of the stages, or alternatively represents the time of occurrence of the trailing transition of the variable width pulse.

FIG. 2 shows a logic circuit implementation of a single bit stage 90 of one embodiment of the invention. As shown, the single bit stage 90 includes a pair of

AND gates

92 and 94, an OR gate 96, another AND gate 98, a switching transistor 100, a capacitor 102, another switching transistor 104, and an inverter 106. Operation of the single bit stage 90 of FIG. 2 will be described with reference to the timing diagram of FIG. 3. At the beginning of a variable width pulse interval, a precharge pulse 108 is applied to the gate electrode of transistor 104, turning this transistor on for the pulse time of pulse 108 and discharging storage capacitor 102 to a source of reference potential, ground in this example. This initializes the single bit stage 90. If the digital data bit D₁ applied to the single bit stage is a logic zero, it will disable AND gate 94 and enable AND gate 92 via inverter 106. On the application of a pulse 112 of clock φA1 to the other input of AND gate 92, the output from AND gate 92 will go high, providing a high input to OR gate 96. In turn, the output signal from OR gate 96 will go high and be applied to one input of AND gate 98. If the start signal 110 is high at this time, AND gate 98 is enable, and its output signal will change from a low to a high level immediately upon the application of the high level input from OR gate 96. The high level output signal from AND gate 98 is applied to the gate electrode of transistor 100, conditioning it to charge capacitor 102 toward +V_S and causing the output voltage V_D to go high coincident with the positive transitions of the φA1 pulse (see waveform V_D (D1=0).

If D₁ is a logic high level, that is a digital "1", inverter 106 will apply a "0" to one input of AND gate 92, disabling that AND gate, and a digital "1" is applied to one input of AND gate 94, enabling that AND gate. When clock pulse signal φB1 changes state to a digital "1", the output signal from AND gate 94 will change state from "0" to "1", which signal is coupled through OR gate 96 to one input of AND gate 98. Since AND gate 98 is already enabled via the high level start signal 110 connected to its other input terminal, the output signal from AND gate 98 will change from a "0" to a "1" state. The "1" stage conditions transistor 100 to provide a logic high output V_D coincident with the positive transitions of the φB1, clock pulse (see waveforms VD(D1=1)).

In the FIG. 2 system the variable width output pulse starts with the leading edge of the precharge pulse and terminates at the leading edge of either the φA1 or φB1 clock pulse. Alternatively the pulse may be considered as having a leading edge defined by the leading edge of either the φA1 or φB1 clock pulse, and a trailing edge defined by the leading edge of the precharge pulse. For typical LCD scanner applications the former definition of the variable clock pulse obtains.

The output signal V_D from a single bit circuit 90 can be applied to the start input of a succeeding stage connected in cascade therewith, for arming or enabling the succeeding stage 90 in the chain, as will be described below. Prior to applying a new data bit to the stage 90, another precharge pulse 108 is applied to the gate of transistor 104 to discharge capacitor 102, and reinitialize the stage 90 for the next bit cycle.

FIG. 4 illustrates the interconnection of a plurality of the single bit stages shown in FIG. 2, for processing or decoding a digital signal containing "n" bits. Note that a precharge signal line 120 is connected in common to the gate electrodes of the transistors 104 of each one of the single bit stages 90. Thus all stages are initialized simultaneously. Also note that two different clock pulses are required for each of the stages 90, with no one stage receiving the same clock pulses as another stage.

FIG. 5 illustrates a timing chart for the example where two single bit stages are connected in cascade for decoding a data signal having two bits D1, and D2. In FIG. 5,

waveforms

108, 110, 112, and 114 are substantially similar to those of FIG. 3, and are associated with decoding the most significant bit of the digital data signal D1.

Timing pulse waveforms

120 and 122 are associated with the clock pulses φA2 and φB2, respectively, for application to the second stage 90 for decoding the least significant bit D2, in this example.

The decoding of the most significant bit D₁ is substantially the same as the decoding process previously described for the single bit stage of FIG. 2. If D₁ is "0", node 130 of the first stage will go "high" responsive to the leading edge of the first occurring clock pulse φA1 shown as pulse 112 in FIG. 5. When node 130 goes "high", AND gate 98 of the next comparator stage 90 is enabled. Now if data bit D₂ is "0", the output signal V_D will go "high" upon the occurrence of the next φA2 clock pulse shown as pulse 120 in FIG. 5. The change in state of the output signal V_D is indicated by waveform 124. This change in state occurs delayed in time by T₀,0 from the positive transition of the START signal. Alternatively, if data bit D₂ is "high" or "1", the output signal V_D will go high coincident with the next occurring clock pulse φB2, as shown by waveform 126 which is delayed by T₀,1 indicated.

If D₁ is "1", the second stage AND gate 98 will be enabled upon the first occurrence (114) of the φB1 clock pulse after the positive transition of the start pulse 110. Assuming D₂ is a "0" the output V_D will go high upon the occurrence of the next φA2 clock pulse (120) as shown by waveform 128 which is delayed by T₁,0. However, if D₂ is a "1", AND gate 92 of the second stage will be inhibited, and AND gate 94 enabled. As a result, transistor 100 of that stage will not be turned on until the occurrence of the next occurring φB2 clock pulse (122) with the output voltage transition being shown as waveform 130 which is delayed by T₁,1.

The four possible pulse width variation for a two bit signal and the particular clock signals φAn, φBn shown in FIG. 5 are illustrated by the waveforms 124-130. Note however that the transition points may be altered by changing the points in time at which clock transitions occur. Note further that once any stage provides a logic high output (at respective nodes 130), the output potential of that stage will not change regardless of changes in either data or clock states. This results because the output state is stored on the respective capacitors 102, which may only be discharged by transistors 104 when strobed by the precharge pulses φpc. Note further that the combination of

transistors

100, 104 and capacitor 102 serves the function of a SET/RESET flip flop with φpc providing the set signal and AND gate 98 providing the reset signal. Thus where practical a bistable element may be substituted for transistors 100,104 and capacitor 102.

The φA1 and φB1 clock cycles illustrated in FIG. 5, have a period T. The clock cycles φA2 and φB2, for the second or next succeeding cascaded stage 90, have a period of one-half T. Such a progression of reduced time periods by a factor of one-half for the clock cycles of one stage relative to a preceding stage is true in this example for any number of cascaded stages 90. Accordingly, if n=4, for decoding a four bit data signal having bits D1, D₂, D₃, and D₄, the timing chart would be as shown in FIG. 6. As shown, for decoding a four bit data word, four single bit stages 90 must be cascaded, requiring four different sets of clock pulses φA1 through φA4, and φB1 through φB4. Considering a digital data signal having a digital four bit value of "0000", the output voltage V_D does not change level from a zero volt to a higher level until a delay time T_d after the occurrence of the first φA1 clock pulse 112, as shown. This corresponds to T₀,0 in the two stage cascade of FIG. 5. The change in level occurs substantially at the initiation of the first occurring φA4 clock pulse 136, as shown.

FIG. 7 illustrates, another embodiment of a one-bit count circuit implemented in dynamic or pulsed logic. In this embodiment, a bootstrapping method is utilized to obtain sufficiently fast switching rates from slow pulsed logic devices such as amorphous silicon devices, permitting the circuit 140 to be used in the system of FIG. 1, for example. Each of the one-bit count stages 140 can be cascaded for decoding multiple data bit words, as previously indicated for the count stages 90. In this example, each count stage 140 includes transistors 142 through 152 of the same conductivity type; boost

capacitors

154 and 156;

interelectrode capacitors

158, 160, 162, 164, 166, and 168 shown in phantom; and stray or

parasitic capacitances

170, 172, 174, and 176, also shown in phantom.

First consider

transistors

144, 148 having their respective source electrodes coupled to node 188 and transistor 145 having its drain electrode coupled to node 188. The source electrode of transistor 145 is coupled to ground potential. The drain electrodes of

transistors

144 and 148 are respectively coupled via

coupling capacitors

154 and 156 to clock busses φAn and φBn. Let the logic input values applied to the gate electrodes of

transistor

144, 145 and 148 be Dn, MI and Dn, respectively. The logic state of node 188 can be represented

 NODE 188=(((Dn·φAn)+(Dn·φBn))·MI).

The signal MI corresponds to the start pulse, but is inverted in polarity relative to the start pulse discussed with respect to FIGS. 2 and 3. As long as the start pulse MI is high the output at node 188 will be low. Conversely, if MI is low and either Dn is high and φAn occurs or Dn is high and φBn occurs, node 188 will exhibit a logic "1" on the occurrence of φBn or φAn. The output potential at node 188 is stored on the stray capacitance 172.

Input signal to the gate electrodes of

transistors

144 and 148 and output signal from node 188 are provided by dynamic, precharge type buffer invertors. In FIG. 7 these buffer invertors include respective pairs (142, 143), (150, 151) and (152, 149) of transistors having their source-drain conduction paths coupled in series between relatively positive +V_s and relatively negative supply potentials. Output signal from the buffer is taken from the interconnection of the pair of transistors. Input signal is applied to the gate electrode of the transistor coupled to the relatively negative supply potential and a precharge pulse is applied to the gate electrode of the transistor coupled to the relatively positive supply potential. The precharge pulse φpc occurs for a relatively short duration at the beginning of each bit period. (Note a bit period for an LCD video display application is a horizontal line time.) The data logic level applied to the signal input of the buffer inverter must be established prior to the termination of the precharge pulse. Refer to the buffer

inverter comprising transistors

142 and 143 which generates the logic signal Dn applied to the gate electrode of transistor 144. The complement of Dn, Dn provided by for example storage circuitry (not shown) is applied to the gate electrode of transistor 143. The precharge pulse φpc is applied to the gate electrode of transistor 142, and the logic signal Dn is available at node 196. If the input signal, e.g., Dn, is at a logic low state, rendering the pull-down transistor (143) nonconductive, the pull-up transistor (142) will charge the output node (196) to the positive supply potential Vs during the occurrence of the φpc pulse. At the termination of the precharge pulse the pull-up transistor (142) is rendered nonconductive leaving the potential Vs stored on the stray capacitance (170) associated with the output node of the buffer inverter.

Conversely, if the input signal (Dn) is a logic high, the pull-down transistor (143) will be conducting, precluding the storage of any charge on the stray capacitance (170) associated with the buffer inverter output node (196). In this instance, at least shortly after the termination of the precharge pulse, the potential at the output node (196) of the inverter will be a logic low. Whether Dn is a logic high or low, the logic value established on capacitor 170 is retained for a data word period, e.g., approximately the active portion of a horizontal line time for the LCD display application.

Precharge buffer invertors are used to apply the data bit values to the

transistors

144 and 148 so that when a logic high level is applied to a respective gate electrode (144, 148), the impedance looking into the source of the data bit value is extremely high. This permits a capacitive voltage boost to the gate electrode as will be described hereinafter. Using a precharge buffer inverter wherein a charge is first stored on the output node of the inverter and then discharged according to the logic level of the applied data, obviates constructing the inverter with ratioed transistors. This permits relatively fast pull-down with relatively small pull-down transistors.

Regarding the input buffer invertors comprising transistors (142, 143) and (150,151) the relatively negative supply potential applied thereto is assumed to be ground potential. With respect to the output

buffer comprising transistors

152 and 149, the relatively negative supply potential is nominally ground potential, however it may be desirable to establish this potential at an amplitude slightly less than the turn on or threshold potential of the transistors 145. The reason for this is as follows. At the beginning of a bit cycle, the output node 190 is precharged to the positive supply potential Vs. This potential is stored on a relatively small stray capacitance 176. If the capacitor 176 is inadvertently discharged, it cannot be recharged (in this system) until the next bit cycle. Therefore, it is imperative that the pull-down transistor 149 be not inadvertently rendered conductive. Raising the potential applied to the source electrode of the transistor 149 raises the potential level which must be applied to its gate before it will turn on. Thus by applying the relatively more positive potential V_B to the source electrode of transistor 149, the noise immunity of the system is increased. The amplitude of the supply V.sub. B is a value which determines the low level of the output signal MO. Since the output signal MO must be capable of exhibiting a logic low value, the amplitude of V_B must be less than the maximum allowed for a logic low value.

The output signal MO is precharged to a logic "1" level at the beginning of a bit cycle and is discharged to a logic "0" level at the first occurrence of a positive pulse at node 188. This occurrence can only happen after the input signal MI goes low.

The relative timing of the FIG. 7 circuitry is illustrated by the FIG. 8 waveforms.

Referring again to FIG. 7, it is contemplated that all of the pull-up

transistors

142, 144, 148, 150 and 152 are enhancement type, and thus charge their respective output nodes in the relatively slow source follower mode. With respect to the precharge buffer invertors this is of little consequence since precharging will nominally occur during the horizontal blanking intervals. The blanking intervals provide sufficient charging time even for relatively small, low mobility, pull-up transistors.

The charging time of node 188 by

transistors

144 or 148 is another matter. Firstly, the drive voltage applied to the gates of

transistors

144 and 148 is no larger than (φpc-V_T) where φpc is the amplitude of the precharge clock pulse applied to either

transistor

142 or 150 and V_T is the threshold potential of the transistors (which may be in the order of several volts). Secondly, the amount of time available is limited. Consider an active line interval of 53 μsec and 8-bit data samples. The clock period of the clock phases φA8, φB8 is 53/128 μsec or 0.415 μs, which is a relatively short period for charging node 188.

The charging capability of the

transistors

144 and 148 is enhanced by boosting the gate drive potentials. Consider that node 196 exhibits a logic "1" and it is desired to charge node 188 via transistor 144. As is known, the larger the gate-source potential applied to a transistor the larger will be the current conducted thereby and consequently the shorter the charging time of a capacitive load.

The impedance looking into node 196 is substantially capacitive since both of the

transistors

142 and 143 are not conducting (if node 196 exhibits a logic "1"). Consider a positive going clock pulse φA applied to the drain electrode 155 of transistor 144. Because there is a logic "1" potential biasing transistor 144 on, node 188 will begin to charge via the drain-source conduction path. Note, however, that a portion of the clock pulse φA applied to the drain electrode of transistor 144 will couple to its gate electrode via the capacitance 158, thereby enhancing the gate drive potential, and turning the transistor on harder. In addition, as node 188 begins charging, a portion of this potential is coupled back to the gate electrode, via the capacitance 160, enhancing the gate drive still more.

The

capacitances

158, 160 and 170 are tailored relative to each other to a) boost the drive potential on the gate electrode of transistor 144 and enhance its current drive when a logic one is applied to its gate; b) insure that transistor 144 is not inadvertently turned on via coupling of clock potential to its gate electrode when transistor 143 is clamping the gate to ground and c) to insure that sufficient clock potential is not coupled via capacitors 158 and 160 to node 188 to inadvertently turn on transistor 149.

The clock signals φA and φB are coupled to the drain electrodes of the

transistors

144 and 148 via

capacitors

154 and 156, to limit the available current for charging the capacitance associated with node 188. Limiting the current available for conduction permits the use of a smaller pull-down transistor 145. Since the current available from capacitively coupled clock signals is proportional to C dv/dt it is advantageous to lengthen the transition of the clock signals. Therefore for the stages illustrated in FIG. 7 clock pulses having sloping leading edges are utilized, that is the clock signals are arranged to have sawtooth like waveforms as illustrated in FIG. 8. The capacitive coupling of the clock potentials requires that larger amplitude clock signals be utilized. As a consequence the potential at the drain electrode of the

transistor

144 or 148 which is turned off may exceed allowable device parameters. To preclude such potential excess, diode connected transistors 146 and 147 are coupled between the respective drain electrodes and a point of clamping potential. The clamping potential is a threshold potential less than the desired clamp level.

In FIG. 9, various waveforms for a single bit stage 140 when transistor 143 is nonconductive are shown. Node 196 has a voltage level over a period of time T as shown by curve 198. Since node 196 is substantially at +V_S statically, or slightly greater than this level, transistor 144 will be turned on. Note that curve 198 is representative of the voltage at node 196 at times that D_n is "low", and the MI or start input signal is "high". Also, under these conditions, curve 200 shows the small voltage impulses 195 at node 188, curve 202 the voltage (+V_S) at node 190, and the waveform 204 is representative of a clock signal for either the φAn or φBn clocks. If the MI signal goes "low" with other signal conditions remaining the same, then node 188 will undergo a voltage transition as shown by curve 206, and node 196 will have a voltage as shown by curve 208, when the boost drive φAn or φBn is applied in the form of clock signal 204 to the count stage 140. The curve 210 is illustrative of the discharging of node 190 when node 188 goes high. In practice, computer simulations have shown that node 190 does not discharge appreciable until about a time T/2 has elapsed, as shown in FIG. 9 for waveform 210. This characteristic may be utilized for enhancing timing for a plurality of cascaded stages.

In FIG. 9, if the MI signal is "high" when either the φAn or φBn clocks occur, then transistor 145 remains conducting, and node 188 can rise only a moderate amount in voltage magnitude as shown by curve 200, even with transistor 144 remaining conductive due to the previously described gate bootstrapping action provided by interelectrode capacitor 158. The rise in voltage at node 188, at this time (waveform 200), is insufficient to turn on transistor 149, unless the voltage increase exceeds the voltage level equivalent to the sum of the threshold and reference voltages (V_TH +V_B). Accordingly, a single impulsive voltage 200 at node 188 at this time does not cause any discharge of node 190. A number of such sub-threshold impulses may have a cumulative affect, however, and may cause significant discharge, over a long time interval, such as 50.0 μs after termination of the precharge pulse 180. To insure that such false discharging does not occur, it was experimentally determined that the maximum impulse amplitude 195 (waveform 200) must be at least 3.0 volts below the voltage threshold V_TH of transistor 149. Accordingly, if the voltage threshold of transistor 149 is 3.0 volts, then the level of +V_B must equal the maximum impulse voltage that may occur at node 188. A 2.0 volt impulse voltage 200 is typical in practice.

In stage 140 of FIG. 7, assume that the D_n data bit is "low", and Dn is "high", thereby turning on transistor 143. The voltage waveforms associated with various nodes under this condition are shown in FIG. 10. Curve 216 shows the voltage at node 196 as it appears regardless of whether the MI signal is "low" or "high". Curve 212 shows the very small resultant disturb voltage at node 188 if the MI signal is "high". Curve 214 shows the somewhat larger voltage at node 188 if the MI signal is "low". Curve 218 is representative of the voltage at node 190, showing that the MO signal remains at +V_S (voltage at node 190), and curve 220 shows the quite high voltage swing appearing at node 155; this voltage 220 now can approach curve 204 in amplitude since device 144 is now nonconductive. This high voltage swing can tend to turn 144 "on" via coupling through capacitor 158. To prevent this, clamping transistors 146 and 147 are provided to limit the amplitude of the curve 220 relative to curve 204.

With one channel transistor 143 conductive, and the other channel transistor 151 nonconductive, or vice versa, an impulse at node 188 will attain an insignificant level provided that

transistors

144 or 148, respectively, remain turned off during times of occurrence of a φAn or φBn channel clock pulse 204, respectively. Assuming that this is the case,

transistors

143 and 151 must be made large enough from a device standpoint to maintain the voltage at node 196 or at node 222, respectively, to less than a transistor threshold voltage above the voltage appearing at node 188, during the times of occurrence of pulse 204 for either φAn or φBn. In practice, if the period T (FIG. 9 or 10) is equal to 0.7 μs,

transistors

143 and 151 each having channel widths w, equal to 15.0 microns will suffice, assuming that

transistors

144 and 148 each have channel widths w equal to 200.0 microns. Accordingly, in this manner, small data switching devices provide control of much large switching devices. This characteristic is believed unique to the present precharged node, bootstrapped circuit stages 140.

FIG. 11 shows a typical timing diagram for a system including three of the one-bit count stages 140 connected in cascade. As shown in this example, the drive pulses for the φAn clock and φBn clock each are assumed to have a pulse duration of T. Note that a minimum pulse period of 5T/2 occurs from the edge of the start pulse for a cascaded chain of three stages. Data outputs occur in steps of 2T for decoding successively increasing values of data signals, as shown. Note that the timing shown in FIG. 11 allows for some margin, that is, it is designed for noncritical timing. This occurs because the MI input signal for a given stage 140 will start to go low at a time delay of T/2 before the φAn clock signal for that stage starts to increase and T/2 after the end of the φBn clock signal for that stage. With such timing, some overlap of clock signals of successive stages can be tolerated.

In FIG. 12 an example of critical timing is shown. In this example, output decoding steps can occur closer (every 3T/2). Now no margin is provided between the times of the MI signal going low, and the starting to fall φBn clock pulse signal for a given stage 140. The T/2 delay before the rise of φAn is maintained. Note that for a given stage 140 the initial delay time remains at 5T/2.

Although various embodiments of the inventions have been described herein for purposes of illustration, they are not meant to be limiting. Variations and modifications of these embodiments of the invention may occur to those or ordinary skill in the art, which modifications are meant to be covered by the spirit and scope of the appended claims.

Claims

What is claimed is:

1. Apparatus for generating pulses of variable width comprising:

a source of n-bit binary values representing desired pulse widths, n an integer;

clock generating means for generating n ordinally numbered pairs of clock signals φAn, φBn, wherein the phases of a pair of clock signals are displaced relative to each other and relative to other pairs and the clock signals of each ordinally numbered pair of clock signals have at least twice the number of pulses as the next lower numbered pair;

n cascade connected count stages each stage having respective input terminals for receiving a pair of clock signals, φAn, φBn, a bit input terminal coupled to said source for receiving a bit, Dn, of said n-bit binary values, a start input terminal and an output terminal, the start input terminal of each stage being coupled to the output terminal of the preceding stage and the output terminal of the lastmost stage providing output pulses; said count stage providing an output transition responsive to a transition of one of said clock signals φAn, φBn for said bit Dn exhibiting first and second states respectively after an occurrence of a state change at the start input terminal of the respective stage.

2. The apparatus set forth in claim 1 wherein the clock signals φAn exhibit a phase difference of 180 degrees with respect to the corresponding clock signal φBn of a clock signal pair φAn, φBn.

3. The apparatus set forth in claim 1 wherein each of said count stages further includes a precharge means, responsive to a precharge signal, for establishing a predetermined output state at its output terminal.

4. The variable pulse width generator set forth in claim 3 wherein said precharge means includes:

a capacitance coupled between said output terminal and a point of substantially constant potential;

a transistor having a primary conduction path coupled between said output terminal and a point of predetermined supply potential, and a control electrode coupled to receive said precharge signal.

5. The apparatus set forth in claim 3 wherein said precharge means includes a bistable circuit responsive to said precharge signal for conditioning said output terminal to exhibit a first state, and responsive to a transition of said one of said clock signals for conditioning said output terminal to exhibit a second state.

6. The apparatus set forth in claim 1 wherein each of said count stages includes:

first logic means coupled to said clock signals φAn, φBn and said data bit for providing at a terminal thereof a signal according to the logic function

(φAn·Dn)+(φBn·e,ovs/Dn/ );

second logic means responsive to signal from said first logic means and a signal S applied to said start input terminal for providing a signal according to the logic function

S·((φAn·Dn)+(φBn·e,ovs/Dn/ )).

7. The apparatus set forth in claim 6 further including:

a first transistor having a principal conduction path coupled between said output terminal and a first supply potential and a control electrode coupled to said second logic means;

a second transistor having a principal conduction path coupled between said output terminal and a second supply potential, and a control electrode coupled to receive a precharge signal for establishing the output terminal at said second supply potential at the beginning of each variable pulse period.

8. The apparatus set forth in claim 1 wherein each of said count stages includes:

first, second and third transistors each having first, second and control electrodes, the second electrodes of said first and second transistors being coupled to the first electrode of said third transistor, the second and control electrodes of said third transistor being coupled to a first supply potential and said start input terminal respectively;

respective means for coupling the first electrodes of said first and second transistors to said φAn and φBn clock signals;

first means for coupling said data bit Dn to the control electrode of said first transistor; and

second means for coupling the complement of said data bit Dn to the control electrode of said second transistor.

9. The apparatus set forth in claim 8 wherein said first means includes a fourth transistor having a principal conduction path coupled between the control electrode of said first transistor and a second supply potential and having a control electrode coupled to receive a precharge signal at the initiation of each variable pulse; and wherein said second means includes a fifth transistor having a principal conduction path coupled between the control electrode of said second transistor and said second supply potential, and having a control electrode coupled to receive a precharge signal at the initiation of each variable pulse.

10. The apparatus set forth in claim 8 further including:

a fourth transistor having a control electrode coupled to the second electrodes of said first and second transistors, and a principal conduction path coupled between said output terminal and a third supply potential;

a fifth transistor having a principal conduction path coupled between said output terminal and a second supply potential and a control electrode coupled to receive a precharge signal.

11. Apparatus for generating variable width pulses comprising:

a source of n-bit data representing a timing pulse of variable width;

means for generating n pairs of clock signals φAn, φBn, successively higher numbered pairs associated with bits of said n-bit data of lower bit significance, and each successively higher numbered pair of clock signals including a greater number of pulses over predetermined intervals than the immediately lower numbered pair;

a plurality n of ordinally numbered stages coupled for pipelined operation wherein each ordinally numbered stage enables the next higher numbered stage, each stage responsive to a data bit Dn of said n-bit data and associated clock signals φAn, φBn for selecting one of said clock signals in accordance with respective first and second states of said data bit Dn.